plus/minus epsilon

The Death of Cars

18 Apr 2021

Something that's a lot of fun for me is finding convincing visions of the future that are also dystopian. The best example of this I've found recently occurred to me while I was listening to a Tesla earnings call, where Elon Musk justified Tesla's market cap with recurring revenue from the future robo-taxi industry. Musk's theory of the future, and it's dystopian ending, goes like this:

Phase 1

In the near future, self-driving cars will become significantly safer than human-driven ones and this will start a positive feedback loop favoring self-driving cars. Insurance companies would likely be the first actors, offering lower premiums for drivers that utilize self-driving the most, and raising premiums on those that don't. But eventually governments, which would be strongly motivated both by public safety as well as a desire to reduce policing costs, will also start to incentivize using self-driving as often as possible in three ways:

  1. Through direct subsidies on self-driving cars
  2. By increasing the difficulty of obtaining a driver's license and reducing the number of driving classes in public schools
  3. By introducing legislation that directly restricts when and where a car may be human-driven

Phase 2

While Phase 1 is occurring, the robo-taxi industry is still developing. These companies develop an initial customer base in cities, where people are already less likely to own a car, by out-competing traditional taxis like Uber and Lyft solely on price. They leverage this initial scale, expertise, and cash flow to move further and further into the suburbs, serving customers that choose not (or never learned how) to drive and see taxis as a more convenient alternative to owning their own car.

A growing customer base allows robo-taxis to become cheaper because the taxi have a higher useful life. It also makes them more dependable because more taxis means it's more likely there's a taxi nearby you when you want it. Eventually robo-taxis would become cheap and dependable enough that it doesn't make financial or practical sense for almost anyone to buy their own car.

You would also expect for robo-taxi companies to have different demands of their vehicle manufacturers than individual consumers. This includes a reduced size, different interior, higher degree of reliability, as well as better internal and external security. In fact, it's in their interest to demand as many changes from vehicle manufacturers as they can. If they can bifurcate a vehicle manufacturer's market enough that the same processes can't be used for robo-taxis and consumer vehicles, they can deprive consumer vehicles of the advantages of their manufacturing scale. Consumer vehicles may lose the scale that they need to be profitable at all.

It amounts to another positive feedback loop where as robo-taxis become cheaper and more dependable, consumer cars (self-driving or not) become more expensive and unattainable. Cars, which were previously a major consumer asset and a symbol of freedom, become a service that some rich-guy CEO can take away from you at any time -- which leads into phase 3.

Phase 3

If a company sells you a product and you do something weird with it, that's your own problem. But if a company provides you a service and you do something weird with it, that's the company's problem. Which is fundamentally why Internet services have so operational considerations around keeping the service online, effectively moderating content, and managing public opinion and public policy. They're under constant pressure from outside activists, malicious hackers, and often even some of their own employees, all of whom want to disrupt the service or abuse the service's centralized power to accomplish political goals.

On the Internet, this isn't super scary because it mostly just results in inconvenient outages, data leaks that violate your privacy, or a reduced ability to reach your audience if your account is banned. But with taxis, the stakes are much higher:

  • An outage or software bug can disrupt the entire road system.
  • A malicious hacker could hijack the vehicle you're in and abduct or kill you.
  • Being banned by the service could result in being permanently stranded, unable to go further than walking distance from your house.

It's important to note that taxi companies would win their markets by excelling at easily perceived metrics like price and convenience, but would be significantly worse at harder to measure things like privacy, security, failing gracefully, decentralizing control.

Robo-taxi companies, having made themselves essential while also unable to manage the subtler downsides of their business, would end up getting heavily regulated. This regulation would almost certainly further entrench them and give governments unprecedented control over their citizens' movement.